The Numbers On AI: Rising Investment, Few Payoffs Beyond Tech

🔍 Read the full analysis: The Numbers On AI: Rising Investment, Few Payoffs Beyond Tech on ThorstenMeyerAI.com

Before you orderOffer from Amazon

Get smart everyday buys delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Apollo economist Torsten Slok says recent data show no visible financial payoff from artificial intelligence across most large U.S. companies outside technology, even as Meta and other firms commit heavily to computing capacity. His October 24 analysis compared revenue per worker and profit margins at the seven largest technology companies with those at the other 493 S&P 500 firms; it found the former rising while the latter remained broadly flat.

Apollo’s charts put the Magnificent Seven — Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia and Tesla — at roughly $780,000 in inflation-adjusted revenue per employee in the third quarter of 2026. The other 493 S&P 500 companies were around $525,000, in a range they had occupied since 2021. Apollo also showed profit margins for the seven rising to about 30%, while the rest stayed near 10%. These are comparisons between groups, not a direct measurement of the effect of AI on each company.

The report points to broader productivity measures that have not yet recorded an AI-driven acceleration. It cites San Francisco Federal Reserve measures of utilization-adjusted total factor productivity and output per hour. Separately, a survey of nearly 9,000 executives found that 83% reported no productivity impact so far, while the average reported effect was 0.2% a year. The source material does not identify the survey’s full methodology or period, limiting what can be inferred from those figures.

At the same time, major spending plans and commitments are expanding. Meta reported $34.91 billion in capital expenditures in the third quarter and forecast $135 billion to $145 billion in capital spending for 2026. Its October 22 filing listed about $355.73 billion in leases not yet started and $454.52 billion in non-cancelable contractual commitments, plus up to $14.72 billion in contingent cloud purchases. Together, the figures approach $825 billion; they cover future obligations, not an immediate cash outlay, and include some commitments that are not AI-related.

The Gap Between AI Spending and Returns

The contrast matters because companies are making long-term decisions about data centers, chips, cloud capacity and power before broad financial returns are evident in available aggregate measures. Large technology firms are positioned to benefit from selling AI services and infrastructure, while other businesses may be paying for tools without yet seeing a measurable lift in output or margins.

That does not establish that AI has no value for non-tech businesses. Group averages can conceal companies or departments that are benefiting, as well as losses that offset gains elsewhere. But the evidence in Apollo’s analysis challenges claims that AI has already raised productivity across corporate America. For executives, investors and workers, the practical question is whether specific uses can improve performance enough to justify their costs.

Some task-level studies cited in the source material report gains: customer-support workers resolved 14% more issues per hour, rising to 34% for novice workers, while retail experiments found sales changes ranging from zero to 16.3%, depending on the workflow. Those results apply to particular settings and should not be treated as a forecast for every company. They illustrate why a lack of broad gains and measurable local benefits can coexist.

What Apollo’s Comparison Measures

Apollo’s October 24 Daily Spark note, titled “No Signs of AI Payoff Outside the Tech Sector,” used several indicators to make its case. Alongside revenue per worker and margins, it cited Ramp spending data showing that the top 10% of firms accounted for 99.5% of model-serving and inference spending, 99% of neocloud spending and 95.6% of spending on OpenAI and Anthropic. Those figures describe Ramp’s more than 85,000 U.S. business customers, not all American companies. Apollo also said 80.4% of the top 1% of AI spenders were venture-backed, compared with 13.1% of median spenders.

Slok characterized the aggregate results as “zero financial payoff” outside tech. That phrase is an interpretation of the comparisons, not a finding that every non-tech company has received no benefit. A flat average cannot distinguish businesses with gains from those with declines, nor isolate AI’s effect from other factors such as labor costs, interest rates or tariffs.

Meta’s filing provides a view of the spending side, but its commitments are not a direct measure of AI returns. The company’s third-quarter release reported revenue of $64.60 billion, up 26%, and operating cash flow of $32.79 billion. Capital spending exceeded that cash flow, leaving free cash flow at negative $2.12 billion, compared with positive $10.6 billion a year earlier. Meta’s commitments include data centers, energy and network infrastructure, cloud capacity and other items; some are not specifically tied to AI.

““No Signs of AI Payoff Outside the Tech Sector.””

— Torsten Slok, Apollo chief economist, in the October 24 Daily Spark note

What the Aggregate Data Cannot Show

The available figures do not establish how much AI caused the observed revenue, margin or productivity trends. Apollo’s comparisons cover broad company groups, while Ramp’s spending sample is limited to its customers. The executive survey’s methodology and timing are not fully detailed in the supplied material, and the cited task-level experiments may not generalize to other industries or workflows.

It is also unclear how much of Meta’s approximately $825 billion in future commitments will ultimately be used, when the costs will be incurred, or what returns those investments will generate. The total combines different kinds of contractual obligations and leases over multiple years, and is not all AI-related. The source material also gives no final outcome for the prediction of consolidation among neocloud firms.

Whether Returns Catch Up With Costs

The next evidence will come from company filings and operating results showing whether spending on AI infrastructure translates into higher revenue, lower costs or sustained productivity gains. Meta’s 2026 capital-spending range and its future contractual obligations give investors figures to track, but they do not by themselves show whether the investment will pay off.

For companies adopting AI, the clearest near-term test is likely to be at the level of specific tasks: whether a tool improves output, service quality or sales enough to exceed its operating and implementation costs. Apollo’s assessment may change as more firms report results, but the current evidence supports a narrower conclusion: investment is substantial, while broad returns outside tech remain difficult to identify.

Key Questions

What did Apollo mean by “zero financial payoff”?

Apollo economist Torsten Slok used the phrase to describe aggregate indicators showing rising revenue per worker and profit margins for seven large technology companies, while the other 493 S&P 500 firms were broadly flat. It is an interpretation of group-level trends, not proof that every non-tech company has received no benefit from AI.

Does the analysis prove that AI has not improved productivity?

No. The cited broad productivity measures have not shown a clear AI-driven increase, but the analysis does not isolate AI’s effect at every company or workplace. Some specific customer-support and retail experiments reported gains, with results varying by task and setting.

What is included in Meta’s roughly $825 billion figure?

The figure combines about $355.73 billion in leases not yet started, $454.52 billion in non-cancelable contractual commitments and up to $14.72 billion in contingent cloud purchases. These are future obligations across multiple years, not an immediate cash payment, and the source says some are not AI-related.

Where have companies reported measurable AI gains?

The source material cites a customer-support study finding 14% more issues resolved per hour, with a 34% gain for novice workers, and retail experiments with sales changes ranging from zero to 16.3%. These results depend on the workflow and do not establish similar gains for all businesses.

Source: ThorstenMeyerAI.com